The tool provides four characteristic interaction modes with the following operation paths:
- Multi-model selection: Click on the top drop-down menu to check the model (MAX=5), e.g. select both Gemini-1.5 and Llama3-70B.
- Web Search Enhancement: Turn on the "Web Search" switch next to the input box before entering a question and the model will retrieve the latest web information (for time-sensitive questions).
- Visual Q&A: When the Gemini model is selected, clicking on the upload button to attach an image (e.g. recipe recognition), the system automatically parses the visual content
- Temporary key management: Dynamically change the API key via Settings panel, changes take effect instantly without restarting the service.
Special features: The output of DeepSeek R1 is specially optimized to automatically clean up the inference process tags and convert Markdown format to plain text to improve reading fluency.
This answer comes from the articleOpen-Fiesta: an open source tool for chatting with multiple AI macromodels at onceThe































 English
English				 简体中文
简体中文					           日本語
日本語					           Deutsch
Deutsch					           Português do Brasil
Português do Brasil